ICOT FREE-SOFTWARE News Letter No.9

                                                               31 March, 1995



ICOT have been researching and developing of the fifth generation computer technologies in the fifth generation computer project and its follow-on project. ICOT research center will be closed at the end of March, 1995.

Even though ICOT research center will be closed, services for ICOT Free Software will be carried out, including ftp-service and software maintenance as they were.



[Table of Contents]




[Newly Released ICOT Free Software]

To date, we have released 84 programs of ICOT Free Software (IFS). We are now preparing to release 16 further programs as IFS. Since IFS's first release in August 1992, we release 100 programs for two and a half year.

The KLIC system version 2, and programs that have been developed in the follow-on project using KLIC system are included in those 16 programs. By using KLIC system, you can employ the fifth generation technologies on your UNIX workstations.

(1) Newly released ICOT Free Software

The following lists the newly released programs:
Symbol Processing: 2
85. KLIC version 2: A parallel implementation of KL1 for general purpose computers
86. Distributed Pool
<b>Knowledge Representation, Knowledge-base, Constraint: 3
87. Parallel Database Management System: Kappa(KLIC version)
88. Parallel Constraint Logic Programming System: GDCC(KLIC version)
89. Heterogeneous Distributed Cooperative Problem-Solving System: Helios
Problem Solving, Meta-reasoning, Theorem Proving: 3
90. MGTP/G(KL1 version/KLIC version)
91. MGTP/N(Model Distribution)
92. Constraint MGTP(Prolog version/KL1 version/KLIC version)
Experimental Application: 8
93. Multiple Sequence Alignment by Parallel Iterative Improvement Methiod (KLIC version)
94. Multiple Sequence Alignment by Genetic Algorithm
95. Multiple RNA-Sequence Alignment Considering Stem Regions
96. Knowledge Base of Biological Reactions
97. Successive State Splitting of Protein Hidden Markov Network
98. DNA Sequence Analysis using Discriminant Analysis
99. DNA Sequence Analysis using Hidden Markov Model and Genetic Algorithm
100 Legal reasoning system: new HELIC-II



(2) Detailed Description


85 KLIC version 2: A parallel implementation of KL1 for general-purpose computers

[Machine] UNIX machine, etc.
[Environment] UNIX, MS-DOS
[Language] C

[Overview]
KLIC version 2 is a portable and high-performance parallel implementation of a concurrent logic programming language KL1 through translation into the langauge C.

[Features]
KL1 is a concurrent logic programming language designed as the kernel language of the FGCS project. KLIC is a portable and high-performance parallel implementation of KL1 for systems having basic functions of UNIX operating system.

Improvements over the version 1 are as follows.



86 Distributed Pool

[Machine] PIM/UNIX machine
[Environment] PIMOS/UNIX, KLIC
[Language] KL1

[Overview]
Distributed pool provides basic in-memory database features where data are distributed to many processing nodes.

[Features]



87 Parallel Database Management System: Kappa (KLIC version)

[Machine] UNIX machine
[Environment] UNIX, KLIC
[Language] KL1, C

[Overview]
Kappa is a parallel database management system based on a nested relational model, and runs on UNIX machines with KLIC.

[Features]

Nested Relational Model
The data model of Kappa is based on a nested relational model so that complex structured data of knowledge information processing systems or genetic information processing systems can be treated efficiently.
Parallel Processing Depending on Data Placement
As Kappa has a query processing capability at each node, the distribution of relations and the horizontal partition of relations give us inter-node parallelism.
Running on UNIX machines with KLIC
Kappa is written in KL1 and runs on parallel and/or distributed environments of computer systems with basic functions of UNIX operating system using KLIC.



88 Parallel Constraint Logic Programming System: GDCC (KLIC version)

[Machine] UNIX machine
[Environment] UNIX, KLIC
[Language] KL1, C

[Overview]
GDCC (KLIC version) is a parallel constraint logic programming language, which is highly declarative, flexible and efficient, and can deal with various constraints including non-linear polynomial equations.

[Features]
Parallel constraint logic programming system GDCC has various constraint solver libraries and the language processor which translates GDCC programs to KL1. To provide a more flexible and efficient language, GDCC parallelizes both logic language and constraint solvers. This system enables you to describe what (declarative knowledge) without how (procedural knowledge).



89 Heterogeneous Distributed Cooperative Problem-Solving System: Helios

[Machine] UNIX machine
[Environment] UNIX
[Language] C

[Overview]
Helios is a system to support the construction of heterogeneous distributed cooperative problem-solving systems by establishing message passing between various problem solvers such as databases, constraint solvers, etc.

[Features]
Helios defines a problem solver as an agent and attaches cooperation mechanisms to it. These functions enable us to construct a heterogeneous distributed cooperative problem-solving system for solving problems with the cooperation of various agents in a distributed environment. Helios encapsulates a problem solver with a 'capsule' module which absorbs its heterogeneity. Communication between agents is carried out by placing them in an 'environment'. An encapsulated environment also becomes an agent.

Helios provides the capsule description language CAPL, the environment description language ENVL, and agent cooperation mechanisms.



90 MGTP/G (KL1 version/KLIC version)

[Machine] PIM/UNIX machine
[Environment] PIMOS/UNIX, KLIC
[Language] KL1, Prolog

[Overview]
MGTP/G is a theorem prover for non-Horn clauses that deal with ground atoms

[Features]



91 MGTP/N (Model Distribution)(KL1 version/KLIC version)

[Machine] PIM/UNIX machine
[Environment] PIMOS/UNIX, KLIC
[Language] KL1

[Overview]
MGTP/N is a theorem prover for Horn clauses.

[Features]
There are two types of MGTP/N: 'model copying' and 'model sharing'.
The difference is how to keep generated atoms.
With model sharing, each PE has a copy of the model candidates. An advantage of model sharing is that we can minimize inter-PE communication since most time-consuming subsumption tests and conjunctive matchings can be performed independently at each PE.
With model copying, the model candidates are distributed to each PE.
This method can obtain memory scalability and more parallelism than the model sharing method. However, it has the drawback that the communication cost increases since generated atoms need to flow to all PEs for subsumption tests and conjunctive matchings.



92 Constraint MGTP (Prolog version/KL1 version /KLIC version)

[Machine] PIM/UNIX machine
[Environment] PIMOS/UNIX, KLIC
[Language] KL1/Prolog

[Overview]
Extended MGTP to solve finite domain constraint satisfaction problems

[Features]



93 Multiple Sequence Alignment by Parallel Iterative Improvement Method (KLIC version)

[Machine] UNIX machine
[Environment] UNIX, KLIC
[Language] KL1, C

[Overview]
This system partially improves temporary alignment in an iterative way, and can effectively achieve high-quality alignment as a result. (KLIC version)

[Features]
Multiple branches of a search tree in this combinatorial problem are evaluated in parallel by using many processing elements in each iteration. The heuristic method, ``Restricted Partitioning Technique,'' prunes a large number of branches in the search tree and makes it possible to solve the combinatorial problem in a practical amount of time.



94 Multiple Sequence Alignment by Genetic Algorithm

[Machine] PIM/UNIX machine
[Environment] PIMOS/UNIX, KLIC
[Language] KL1

[Overview]
This parallel system, the combination of a genetic algorithm and an iterative alignment algorithm, produces better results than iterative aligners which employ hill-climbing search strategies.

[Features]
This parallel alignment system featuring a genetic algorithm can improve an alignment score as rapidly as that with the hill-climbing algorithm, and the system gradually increased the score to a higher level which might be close to the optimal alignment score. The reason why the genetic algorithm shows such high performance seems to be the modularity of multiple alignment problems. If we can replace a part of a sequence alignment with a better part from another, we obtain a better multiple sequence alignment. Crossover operators achieve this replacement in a statistical manner.



95 Multiple RNA-Sequence Alignment Considering Stem Regions

[Machine] PIM/UNIX machine
[Environment] PIMOS/UNIX, KLIC
[Language] KL1

[Overview]
This system refines a rough alignment using a parallel simulated annealing algorithm taking into account connected base pairs in RNA stem regions.

[Features]
In general, the sum-of-pair alignment does not show complete alignment in terms of stem structures, but merely indicates some possible stem regions. This system refines the rough alignment with a temperature parallel simulated annealing algorithm which optimizes the score obtained by connected base pairs and covariance matches. This procedure gives us a proper RNA-sequence alignment, from which we can specify stem regions.



96 Knowledge Base of Biological Reactions

[Machine] UNIX machine
[Environment] UNIX, Motif, micro-Quixote
[Language] C, g++, Quixote

[Overview]
This system provides illustrations of biological reactions, signal transductions, gene expressions and their relationships using Quixote.

[Features]



97 Successive State Splitting of Protein Hidden Markov Network

[Machine] UNIX machine
[Environment] UNIX, Motif
[Language] C

[Overview]
Successive State Splitting is an optimization mechanism of hidden Markov networks using the same expectation maximization criteria as the Hidden Markov Model (HMM) learning.

[Features]



98 DNA Sequence Analysis using Discriminant Analysis

[Machine] UNIX machine
[Environment] UNIX
[Language] C

[Overview]
This program extracts signal patterns in DNA sequences using a discriminant analysis (class II quantification theory).

[Features]
Extraction of signal patterns is one of important problems in genetic information processing. The patterns are utilized for considering DNA-protein interaction and predicting functional sites in gene identification. Using a discriminant analysis (class II quantification theory), this program extracts signal patterns in DNA sequences consisting of 2 categories (positive and negative samples).



99 DNA Sequence Analysis using Hidden Markov Model and Genetic Algorithm

[Machine] UNIX machine
[Environment] UNIX
[Language] C

[Overview]
This program extracts signal patterns in DNA sequences using Hidden Markov Model (HMM) and Genetic Algorithm (GA).

[Features]
Extraction of signal patterns is one of important problems in genetic information processing. The patterns are utilized for considering DNA-protein interaction and predicting functional sites in gene identification. This program extracts signal patterns in DNA sequences using HMM and GA. The patterns are expressed in the form of HMM. The topology of HMM is optimized by GA evaluating the likelihood of HMM and complexity of the topology. A subroutine for evaluating each HMM is impremented as a process, and executed concurrently.



100 Legal reasoning system: new HELIC-II

[Machine] UNIX machine
[Environment] UNIX
[Language] C

[Overview]
The new HELIC-II is an experimental software tool that simulates the complex reasoning processes of legal experts and is capable of some types of legal reasoning.

[Features]
The new HELIC-II simulates legal reasoning strategies as follows.



To refer detail of these 16 programs, the "ICOT Free Software Catalogue" including all of 100 IFS programs, and the "ICOT Free Software Catalogue (Selected Version)" including 23 IFS programs released in the follow-on project has recently published in December 1994. Anyone wishing to obtain these catalogues is invited to contact the IFS desk via e-mail, mail or fax at the address given at the end of this newsletter.



[Call for KL1 Programs]

There are now many users of KLIC. We at ICOT would like to gather many KL1 programs written by these users into a library for those new to KL1.

There are many KL1 programs in ICOT Free Software, but, none of them are appropriate in size for reading or modifying by KL1 beginners. We would therefore like to ask you to send KL1 programs for those KL1 biginners. We would like to release a KL1 program library with the same conditions as ICOT Free Software. Those who agree with this proposal should send KL1 programs by e-mail to "ICOT Free Software Desk".

We will gather those programs and release them as a program library the same as for ICOT Free Software. We hope that many of you will send Kl1 programs to us suitable for KL1 beginners.



[User's group]

Anyone interested in organizing a user's group to carry out research on any particular item of icot free software, for the purposes of revision or improvement is invited to contact the ifs-desk via e-mail at the address given at the end of this newsletter.

We hope to feature some of your proposals in the next issue of the newsletter. we will also forward reactions to your proposals to you.

Revised programs can be stored on the ftp server at icot, if you feel that your revision would be useful for other users.



[user's group for cu-prolog]

After its release as ICOT Free Software, a constraint logic programming language cu-Prolog, including Prof. Sirai's Macintosh/MS-DOS version, has been used by many students, engineers, and CLP/NLP researchers.

In order to exchange comments and information about cu-Prolog, constraint-based grammar formalisms, and CLP, we have organized a cu-Prolog user's group. The address is


			    cup@icot.or.jp

Please send e-mail to

			cup-request@icot.or.jp

to join the mailing list. Currently, 49 researchers are on the mailing list.



[Contacts]

For information on IFS, access

			    ifs@icot.or.jp

by e-mail. If you receive a paper edition of this newsletter, let us know your e-mail address and we shall send you the electronic edition. If you do not have an e-mail facility, contact the address below. All available IFS is listed in "ICOT Free Software Catalogue", and all IFS released in the follow-on project is listed in "ICOT Free Software Catalogue (Selected Version)". If you do not have a copy of either catalog, supply the IFS-desk with your postal address and we shall arrange to send a copy to you.

If your colleagues or acquaintances are interested in IFS, let us know their name and both their e-mail and postal addresses, and we shall arrange to send them both the newsletter and catalogue.

ICOT Free Software desk Institute for New Generation Computer Technology 21st Floor, Mita Kokusai Bldg. 4-28, Mita 1-chome Minato-ku, Tokyo 108 Japan FAX: +81-3-3456-1618


www-admin@icot.or.jp